Google warns LLMjacking attacks cost enterprises up to $1 million in cloud charges

Threat actors are currently hijacking corporate access to Large Language Models (LLMs), leading to substantial costs for some businesses, with Google alerting that these attacks can inflate cloud charges by as much as one million dollars. This malicious practice, termed LLMjacking, represents a modern iteration of resource hijacking specifically tailored for the age of LLMs. Essentially, an attacker manages to acquire an API key through several vectors, including phishing schemes, vulnerabilities in source code, improperly set up environments, or code left exposed on the client side. Google issued this warning on Monday, detailing how these actors are fraudulently capitalizing on the costly computational power and advanced AI services provided by enterprises.